README.md 1.17 KB
Newer Older
AF3's avatar
AF3 committed
1
2
# Cancer Metastasis using CNNs and transfer learning

AF3's avatar
AF3 committed
3
## Links
AF3's avatar
AF3 committed
4
5
6
7
Link to dataset: https://www.kaggle.com/c/histopathologic-cancer-detection/overview

Dataset is also available at:
```bash
AF3's avatar
AF3 committed
8
cluster.cs.earlham.edu:/eccs/home/afarah18/488/markIII/histopathologic-cancer-detection
AF3's avatar
AF3 committed
9
10
11
12
```

Link to paper: 

AF3's avatar
AF3 committed
13
14
15
16
Link to Video:

Link to Poster:

AF3's avatar
AF3 committed
17
## Software architecture diagram
AF3's avatar
AF3 committed
18
19
![Software architecture diagram](https://portfolios.cs.earlham.edu/wp-content/uploads/2021/03/Screen-Shot-2021-03-22-at-11.42.40-PM.png)

AF3's avatar
AF3 committed
20
## Files
AF3's avatar
AF3 committed
21
22
23
24
25
26
27

basicCNN.py -> This file includes code for training a model with randomly initilized weights and biases

VGGCNN.py -> This file includes code for training a VGG model with ImageNet transfer learning

VGGLessData.py -> This file includes code for training VGG with transfer learning with half of the dataset

AF3's avatar
AF3 committed
28
validate.py -> This file includes code for evaluating the results of the three models
AF3's avatar
AF3 committed
29

AF3's avatar
AF3 committed
30
## How to use
AF3's avatar
AF3 committed
31
32
33
After installing the dependencies each file can be run simply by using:
```bash
python3 <filename>
AF3's avatar
AF3 committed
34
35
36
37
38
39
40
41
42
43
```

Naturally validation file has to be run after all three models have been trained and saved as .h5 files.

## List of dependencies
+ NumPy
+ Pandas
+ TensorFlow
+ Keras
+ MatplotLib